Large-scale data processing may include parallel processing, which generally involves performing some operation over each element of a large data set. The various operations may be chained together in a data-parallel pipeline to create an efficient mechanism for processing a data set.